Cross section fitting method for combined model tree

The invention discloses a cross section fitting method for a combined model tree. The method comprises the following steps of 1), establishing a cross section expression: expanding the cross section expression into a principal term and multiple sub-terms according to cross section state parameter combinations; and 2), performing cross section fitting: performing fitting and pruning on the principal term and the sub-terms by adopting a model tree method, wherein only the fuel consumption is selected as a classifiable variable in a calculation process of the model tree method, and a fitting polynomial contains coupling quadratic terms of state parameters. According to the method, a dependency relationship between the cross section and the state parameters is considered when the cross section expression is established; the fitting and pruning are performed through the model tree method, so that the fitting precision can be ensured, the storage capacity can be reduced, and the generalization capacity can be impro